Contents lists available atScienceDirect
Aquaculture
journal homepage:www.elsevier.com/locate/aquaculture
Dietary selenium required to achieve body homeostasis and attenuate pro- inflammatory responses in Atlantic salmon post-smolt exceeds the present EU legal limit
P. Antony Jesu Prabhu
a,⁎, Elisabeth Holen
a, Marit Espe
a, Marta S. Silva
a, May-Helen Holme
c, Kristin Hamre
a, Erik-Jan Lock
a,b, Rune Waagbø
a,baInstitute of Marine Research, P.O. Box 1870, 5817 Bergen, Norway
bDepartment of Biological Sciences, University of Bergen, P.O. Box 7803, 5020 Bergen, Norway
cSkretting Aquaculture Research Centre, P.O. Box 48, 4001 Stavanger, Norway
A R T I C L E I N F O Keywords:
Selenium Atlantic salmon Requirement Organic selenium Availability Inflamation
A B S T R A C T
Selenium (Se) supplementation either as inorganic or organic form was evaluated in Atlantic salmon post-smolt in vivoandin vitro. The basal diet was formulated to be low in fish meal and contain 0.24 mg Se kg−1; six other diets with Se inclusion of 0.15, 0.4, 0.7 or 1.1 mg kg−1as sodium selenite (SS) and 0.15 or 0.4 mg kg−1as L- selenomethionine (SM) were formulated from the basal diet. The diets were fed to Atlantic salmon post-smolt (mean initial weight, 216 ± 27 g) in triplicate groups (35 fish tank−1) and reared in flow-through seawater (33 ppt) at 10–12 °C for 9 weeks. At the end of the feeding trial, whole fish and tissues were sampled forin vivo assessment; whereas, liver cells and head kidney leukocytes (HKL) were isolated and their primary cultures used forin vitroassessment following exposure to hydrogen peroxide (H2O2), lipopolysaccharide (LPS) or poly I:C (PIC). Growth, feed intake, feed conversion ratio, specific growth rate, hepato-somatic index, proximate com- position and mineral concentration of the whole fish (except for Se) were unaffected by dietary Se (p > .05).
Hematocrit was significantly higher in fish fed the 0.4 mg Se supplemented feeds, irrespective of the Se source (p= .02). The Se concentration in whole body, liver, muscle, plasma, kidney and liver/kidney Se ratio increased with increasing dietary Se concentration (p < .0001). Level of oxidised glutathione (GSSG) in liver and head kidney followed a quadratic function (p< .05) indicating that the concentrations were lower at intermediate SS supplementation of 0.4 and 0.65 mg Se kg−1(total Se, 0.65 and 0.87 mg Se kg−1). Impact of Se sources on glutathione redox status was similar. Slope-ratio analysis revealed SM to be more efficient than SS in improving apparent availability, whole body or tissues Se status, Se retention and reducing Se loss to the environment.In vitro, the mRNA expression ofp38mapkandaif, in liver cells were affected by the impact of dietary Se, but not by the treatment of H2O2(p < .05). In the HKL, the LPS and PIC induced pro-inflamatory action ofil-1β,cox2,nfkβ andviperinwere attenuated by SM supplementation, but not by SS (p < .001). Dietary Se supplementation required to the basal diet containing 0.24 mg Se kg−1was 0.41 mg Se as SS (total 0.65 mg kg−1) or 0.17 mg Se as SM (total 0.41 mg kg−1) based on body Se homeostasis or tissue Se status. SM inclusion at 0.4 mg kg−1diet (total, 0.65 mg kg−1) attenuated LPS or PIC induced pro-inflammatory responsesin vitro. Overall, Se require- ment of Atlantic salmon post smolt was 0.27 mg kg−1diet, on availale basis. Dietary Se level required to maintain body Se homeostasis and improved health status of Atlantic salmon fed plant-based diets (0.65 mg kg−1diet) exceed the existing EU maximum limit of 0.5 mg Se kg−1diet. SM as the Se source in salmon feeds has the potential to improve salmon health and reduce Se emissions from Norwegan salmon farming by 60 to 70%.
https://doi.org/10.1016/j.aquaculture.2020.735413
Received 24 October 2019; Received in revised form 23 April 2020; Accepted 24 April 2020
Abbreviations:SS, sodium selenite.; SM, selenomethionine.; LPS, lipopolysaccharide.; PIC, polyinosinic:polycytidylic acid.; AAC, apparent availability coefficient.;
il-1b, interleukin 1 beta.; p38mapk, p38 mitogen-activated protein kinases.; 5lox, Arachidonate 5-lipoxygenase.; cox2, cyclooxygenase2.; sod2, superoxide dis- mutase2.; casp3, caspase 3.; cat, catalase.; viperin, virus inhibitory protein, endoplasmic reticulum [ER]-associated, interferon [IN]-inducible.; bcl2, B-cell lymphoma 2.; tnfα, tumor necrosis factor alpha.; nfkβ, nuclear factor kappa-beta.; pparα, peroxisome proliferator activated receptor alpha.; gpx2, glutathione peroxidase 2.;
gpx3, glutathione peroxidase 3.; aif, apoptosis-inducing factor.; β-actin, beta actin.; ef1α, elogation factor1 alpha.; rpl13, ribosomal protein L13.
⁎Corresponding author.
E-mail addresses:[email protected](P. Antony Jesu Prabhu),[email protected](E. Holen),[email protected](M. Espe),[email protected](M.S. Silva), [email protected](M.-H. Holme),[email protected](K. Hamre),[email protected](E.-J. Lock),[email protected](R. Waagbø).
Aquaculture 526 (2020) 735413
Available online 25 April 2020
0044-8486/ © 2020 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
T
1. Introduction
Selenium (Se) is an essential nutrient exerting its role as seleno- proteins (SeCys) and selenium containing proteins (SeMet) in numerous biological processes involving free radical and antioxidant metabolism, immune and inflammatory responses (Regina, 2015). Diet is the major route of Se supply to fish (Janz, 2012) and the dietary Se level essential to satisfy the requirement to fish ranges between 0.15 and 1.85 mg kg−1diet, across different fish species (Antony Jesu Prabhu et al., 2016). Among salmonids, rainbow trout has a minimal require- ment in the range of 0.15–0.38 mg Se kg−1diet (Hodson and Hilton, 1983); whereas, the Se requirement is yet to be established in Atlantic salmon (NRC, 2011). An indication of a dietary Se range (0.6 to 0.9 mg kg−1diet) to satisfy requirement in post-smolt Atlantic salmon was obtained recently from a multi-nutrient regression study (Antony Jesu Prabhu et al., 2019). Over the years, salmonid feeds included considerable levels of fish meal (Aas et al., 2019), a good source of Se (Julshamn et al., 1978). The fish meal based salmonid feeds often contain close to 1 mg Se kg−1diet and therefore Se supplementation was not considered essential to meet the minimal Se requirements (Bell and Cowey, 1989;Lorentzen et al., 1994). Reduction in fish meal with concomitant increase in plant protein sources has reduced Se supply (Betancor et al., 2016;Fontagne-Dicharry et al., 2015;Sissener et al., 2013), altered Se speciation (Godin et al., 2015;Sele et al., 2018) and reduced dietary availability (Antony Jesu Prabhu et al., 2014; Silva et al., 2019). Nevertheless, the maximal permitted Se concentrations in animal feeds, including fish feeds in the EU is 0.5 mg Se kg−1 in complete feed (European Commission, 2014). Therefore, the im- portance of Se supplementation in plant ingredient based fish feeds gained significance ever more and with it, the bioavailability of dif- ferent Se forms.
The chemical form of Se differentially affects the bioavailability and toxicity in fish feeds (Berntssen et al., 2017). Different inorganic (so- dium selenite and sodium selenate) or organic Se forms (seleno- methionine, selenocysteine, selenised yeast) have been studied in sal- monids and found to vary in their bioavailability and toxicity thresholds (Bell and Cowey, 1989; Berntssen et al., 2018; Berntssen et al., 2017; Fontagne-Dicharry et al., 2015; Hamilton et al., 1990;
Hardy et al., 2010;Hilton et al., 1980, 1982;Hodson and Hilton, 1983;
Pacitti et al., 2016;Pacitti et al., 2015;Rider et al., 2010;Silva et al., 2019). A common consensus from the above listed studies is that, compared to inorganic Se, the organic Se forms show greater avail- ability and higher ability to increase the Se status of the fish (Antony Jesu Prabhu et al., 2016).
Selenium, through selenoproteins influence innate and acquired immune responses, especially under conditions of stress (Arthur et al., 2003;Turner and Finch, 1991). During stress, Se stored as SeMet in Se- conatining proteins is liberated by transselenation via the transsul- furation pathway for the synthesis of selenoproteins (Burk and Hill, 2015). In accordance, organic Se is suggested to have a better ability to protect fish against physical, chemical or pathogen-induced stressors (Khan et al., 2017). In mammals, Se mediates inflammation by mod- ifying cellular redox tone, which in turn regulates thenfkb, mapkand tnfα signalling pathways, affectingcoxandloxenzymes (Bonizzi and Karin, 2004; Mattmiller et al., 2013). These pathways also regulate inflammation in fish and are responsive to dietary Se levels and sources, in vivo(Pacitti et al., 2016;Zheng et al., 2018a;Zheng et al., 2018b).
Further, the stress mitigating effect of organic Se was observed often at supra-nutritional levels exceeding the minimal dietary requirement (Küçükbay et al., 2009;Pacitti et al., 2016;Rider et al., 2009;Wang et al., 1997; Zheng et al., 2018a; Zheng et al., 2018b). On the other hand, excess supply of dietary Se will increase load to the environment (Tornero and Hanke, 2016). The key to minimize Se load to environ- ment is by improving dietary Se availability and retention. Therefore, it is important to improve the availability of dietary Se for better fish health, minimize environmental impact and to comply with the
legislation.
Building on the information on the dietary Se level to satisfy the requirement with a multi-nutrient design (Antony Jesu Prabhu et al., 2019), this study was a targeted approach designed to determine dietary Se requirement and relative efficiency of two Se sources in Atlantic salmon post-smolt. Classical in vivo responses were com- plemented with a pathogen-mimic challengein vitroto assess Se bioa- vailability and requirement in Atlantic salmon post-smolt fed diets low in fish meal.
2. Materials and methods
2.1. In vivo study 2.1.1. Experimental design
The study followed a regression design with 5 graded levels of dietary Se supplemented as sodium selenite (SS) to determine the dietary Se level in plant-based diets at which Se requirement is satisfied in Atlantic salmon post-smolt. Additionally, to assess the relative bioavailability of an organic Se source namely selenomethionine (SM), two diets with graded inclusion of SM were tested. The former was assessed through linear and non-linear regression models, while the latter using slope-ratio method.
2.1.2. Diet formulation and selenium levels
The basal mix was formulated to be low in Se concentration and thus the fish meal inclusion level was considerably reduced compared to the formulation used in commercial Atlantic salmon post-smolt feeds (Table 1). The selenium concentration of the major ingredients used in the basal mix was analysed from three different batches prior to feed production to ensure low basal Se levels (Table 2). The basal mix was supplemented with Se either as (i) sodium selenite (SS; Na2SeO3, 4.5%
Se, DSM nutritional products, Basel, Switzerland) or (ii) as L-seleno- methionine (SM; C5H11NO2Se, 0.16% Se, Orffa Additives, Werkendam, The Netherlands) to formulate seven experimental diets. The analysed Se concentrations in the experimental feeds were 0.24 (basal diet), 0.34, 0.65, 0.87 and 1.4 mg kg−1in SS diets, 0.38 and 0.63 mg kg−1in SM diets (Table 3). The Se levels were selected based on the range of dietary Se level at which tissue saturation occured in Atlantic salmon post-smolt, 0.6 to 0.9 mg kg−1diet following the inclusion of a multi- nutrient package, including Se (Antony Jesu Prabhu et al., 2019). The actual Se concentration in the basal diet (0.24 mg kg−1) was higher Table 1
Formulation and composition of the basal experimental diet.
Ingredients (%)
Whole wheat 10.0
Corn gluten 10.0
Wheat gluten 20.0
Soy protein concentrate 30.0
Faba beans, whole 1.0
Fish meal⁎ 1.0
Fish oil⁎ 10.5
Rapeseed oil† 13.0
Micro-ingredients and Se-free premixes‡ 4.4
Yttrium premix¥ 0.1
Proximate composition (analysed,n= 7)
Dry weight (%) 93.1 ± 0.4
Energy (kJ/g) 23.9 ± 0.2
Lipid (%) 25.5 ± 0.5
Protein, analysed as N * 6.25 (%) 44.0 ± 0.1
Ash (%) 3.8 ± 0.1
Phosphorus (%) 0.82 ± 0.02
* North-Atlantic. † European, non-GM. ‡ Contains monoamonium phosphate, histidine HCl,L-lysine and DL-methionine and astaxanthin; standard vitamin and mineral mix, excluding selenium (Se). ¥ contains 10% yttrium oxide as the inter marker for apparent availability estimation.
than the desired level of 0.1 mg kg−1possibly due to high variations in Se concentration of the ingredients between batches(Table 2). Conse- quently, the Se concentrations in the experimental feeds were also higher than desired (Table 3). The mean analysed protein, lipid, ash and energy content of the diets were respectively 44%, 26%, 3.8% and 24 kJ g−1. All the diets contained yttrium oxide at 0.01% as inert di- gestibility marker. The diets were extruded and the pellet size was 4.0 mm (Skretting ARC).
2.1.3. Fish, feeding and rearing
Seven hundred and thirty-five Atlantic salmon post-smolt (mean weight, 216 ± 27 g), were distributed among 21 tanks with 35 fish in each (mean biomass, 7.56 ± 0.04 kg). The seven experimental diets were randomly assigned to the 21 tanks in triplicates. Prior to experi- mental feeding, the fish were maintained on commercial feeds (Spirit V 75-50A, 3.0 mm, Skretting). The fish were reared in circular tanks (volume, 1 m3) supplied with flow-through seawater (salinity, 33 ppt) at 10–12 °C. The fish were fed the experimental diets twice a day to apparent satiation for 9 weeks. The uneaten feed pellets were collected to estimate the actual feed intake.
2.1.4. Fish sampling and analytical methods
The experiment was approved by the Norwegian Food Safety Authority (Mattilsynet) and conducted in accordance with the Norwegian laws and regulations concerning experiments with live an- imals. The feeding trial lasted 9 weeks and the all fish were individually weighed and measured for total length at the start and the end of the trial. Twenty fish were sampled at the start of the trial and homo- genized for whole body proximate and mineral composition analysis. At the end of 9-week experimental feeding, all the fish were euthanized with an overdose (6 ml L−1) of tricaine methanesulphonate (MS-222;
PharmaQ, Bergen, Norway). All the fish were stripped and the collected faeces sample was pooled per tank, freeze dried for 72 h at −80 °C and stored at room temperature until further analysis (Silva et al., 2019).
Ten fish per tank were then evacuated for any residual gut content and homogenized for whole body analysis. Further, blood samples were collected from six fish per tank using heparinized vacutainers and the plasma separated after centrifugation for 5 min at 3000gand the re- covered plasma was collected as individual samples. Hematocrit was measured from a subsample of the blood (Antony Jesu Prabhu et al., 2017). After sampling for blood, the fish were dissected and sampled for liver, kidney and muscle, each tissue was pooled and homogenized per tank. All the tissue samples were stored at −20 °C until analysis. In- dividual samples of liver and head kidney from three fish per tank were flash frozen in liquid nitrogen and subsequently stored at −80 °C for the analysis of GSH, GSSG and calculation of the associated redox po- tential (Hamre et al., 2016). The diets were homogenized and analysed for determination of dry matter, ash, lipid, protein following standard procedures. Briefly, dry matter was measured after drying at 103 °C for 24 h; ash content determined by combustion in a muffle furnace at 550 °C for 16–18 h; lipid was determined following ethyl-acetate and acid-extraction in fish tissue and feeds, respectively (Lie, 1991); protein (6.25 x nitrogen) was measured with a nitrogen analyser (Vario Macro Cube, Elementar Analysensysteme GmbH, Germany) according to AOAC official methods of analysis (Sweeney and Rexroad, 1987). Se- lenium analysis in diets, faeces and tissue samples; and yttrium in feed and faeces samples were performed by inductively coupled plasma mass spectrometry (ICP-MS) (Silva et al., 2019).
2.1.5. Calculations
Apparent availability coefficient, % = 100 - [100*(yttrium in feed/
yttrium in faeces)*(selenium in faeces/selenium in feed)].
Selenium retention, % = [(final biomass * final Se content) - (initial biomass * initial Se content)] * 100/ (Total feed intake * Se content in feed).please correct as required.
Faecal selenium loss, % = 100 – apparent availability coefficient.
Non-faecal selenium loss, % = 100*[(dietary Se intake - body Se gain – faecal Se loss)/dietary Se intake].
Body Se homeostasis = Final Se concentration – initial Se con- centration.
Environmental load calculations: The retention and loss were ex- trapolated to estimate the Se load from Norwegian salmon farming assuming the annual Atlantic salmon production and feed consumption data (Aas et al., 2019). In addition, the median concentration of Se in Norwegian salmon feeds from the Norwegian fish feed surveillance program (Sanden et al., 2017) were used to estimate the intake, re- tention and loss considering a ‘business-as-usual’ scenario. The quantity of Se retained or lost to the environment was then compared with the three different scenarios of dietary Se found to satisfy requirement in this study, SS (0.4 ppm supplementation), SM (0.15 ppm supple- mentation) and SM (0.4 ppm supplementation).
Glutathione redox potential: The two-electron half-cell reduction potential of the 2GSH/GSSG redox-couple was calculated according to the Nernst equation,
Eh = E0′- RT/nF ln([GSH]2/[GSSG]).where the GSH and GSSG Table 2
Selenium concentration of feed ingredients used in Atlantic salmon feeds.
Ingredient Analysed Se concentration (mg kg−1) Mean ± SD Batch 1a Batch 2b Batch 3c
Fish meal, North
Atlantic 2.72 2.48 3.12 2.77 ± 0.31
Whole wheat 0.02 0.16 0.11 0.10 ± 0.07
Wheat gluten 0.19 0.08 0.12 0.13 ± 0.06
Corn gluten 0.12 0.15 0.16 0.14 ± 0.02
Faba bean, dehulled 0.08 0.03 0.04 0.05 ± 0.03
Hi-Pro Soya 0.04 0.01 0.09 0.05 ± 0.04
Soy protein concentrate 0.03 0.04 0.18 0.08 ± 0.08 In every batch, two samples were analysed (n = 2).
a Analysed June 2017.
b Analysed Dec. 2017.
c Analysed April 2018.
Table 3
Study design, selenium (Se) supplementation and analysed concentration of Se in the experimental diets.
Se source Diet code Se inclusion (mg kg−1diet) Desired Se levels (mg kg−1diet) Rationale and reference to Se levels tested Analysed Se levels (mg kg−1diet)
Basal diet A 0 0.1 Low1 0.24
SS (Na2SeO3) B 0.15 0.25 Sub-optimal1 0.34
C 0.4 0.5 Requirement range1; Max limit2 0.65
D 0.7 0.8 Requirement range1 0.87
E 1.1 1.2 Above requirement1 1.4
SM (Se-Met) F 0.15 0.25 Low1 0.38
G 0.4 0.5 Requirement range1; Max limit2 0.63
1Dietary total Se level needed for tissue saturation of Atlantic salmon post-smolt in seawater, 0.6 to 0.9 mg kg−1diet (Antony Jesu Prabhu et al., 2019).2Maximum permitted concentration for total Se in complete fish feeds with 12% moisture is 0.5 mg kg−1diet (European Commission, 2014). SS, sodium selenite, Na2SeO3, 4.5%
Se, DSM nutritional products, Basel, Switzerland : SM, L-selenomethionine, C5H11NO2Se, 0.16% Se, Orffa Additives, Werkendam, The Netherlands.
concentrations are in M and Ehis given in volts. E0′ is the standard reduction potential at pH 7 and 25 °C and was assumed to be −0.240 V (Kemp et al., 2008;Schafer and Buettner, 2001).
2.2. In vitro experiments
One week prior to the final sampling, fish from five treatment groups namely basal diet (A), two SS groups (B and C) and two SM groups (F and G) were sampled for isolation of liver cells (n= 5) and head kidney leucocytesin vitro(n= 3). Tissue samples were also col- lected from these fish to study gene expression changesin vivo. The cells were isolated as described originally in Espe and Holen (2013)fol- lowing modifications as perMartins et al. (2019).
2.2.1. Isolation and preparation of liver cells
Briefly, the fish were anesthetized in MS-222 (0.5 ml L−1), liver perfused clean with a 0.09 M HEPES buffer (1.4 M NaCl, 0.067 M KCl and 0.03 M EDTA, pH 7.4), at a flow rate of 4 ml min−1. The liver was digested in 0.1% type IV collagenase (prepared in 0.9 M HEPES per- fusion buffer), the cells harvested in 10 ml of 10% PBS buffer (0.002 M KH2PO4, 0.02 M Na2HPO4, 0.03 M KCl and 0.14 M NaCl, pH 7·4), centrifuged at 50gfor 5 min, filtered (sterile, 100 mm filter) and wa- shed twice in the perfusion buffer, resuspended in Leibovits-15 medium supplemented with 10% fetal bovine serum (FBS, BioWhittaker), 1%
glutamax (Gibco), 1% antibiotic Antimycotic (Penicillin-Streptomycin 50 U ml−1, BioWhittaker) and assessed for viability with the Trypan Blue exclusion test (Lonzo; Medprobe).
2.2.2. Isolation and preparation of head kidney leucocytes
In brief, the head kidneys from each fish was aspirated, added to sterile isolation buffer (150 mM NaCl and 24 mM EDTA, pH 7.2) and then squeezed through a 40 μM Falcon cell strainer to 50 mL tubes. The cell filtrate was washed by centrifugation (400gfor 5 min, at 4 °C). The resulting cell pellet was re-suspended in the isolation buffer and layered carefully on top of equal amounts of diluted Percoll at a density of 1.08 g ml−1. The tubes were then centrifuged at 800gfor 5 min, at room temperature. The cell layer in the interface was collected and the cells were pelleted by centrifugation (400gfor 5 min, at 4 °C). The cell pellet was then re-suspended in Leibovits-15 medium for culture and use in challenge experiments.
2.2.3. Primary cell culture and challenge
The isolated liver cells and head kidney leucocytes were culturedin vitro(Martins et al., 2019). The cells were cultured on 6 well tissue culture plates in 2 ml of L15 medium for 24 h, incubated at 9 °C.
Subsequently, on day two, the liver cells were challenged with a pro- oxidant, 200 μM H2O2(Espe et al., 2015) and the head kidney leuco- cytes with 100 μg ml−1 lipopolysaccharide (LPS, Sigma-Aldrich) or 50 μg ml−1polyinosinic:polycytidylic acid (Poly I:C or PIC, Sigma-Al- drich) (Martins et al., 2019). The LPS is a gram-negative bacterial cell wall component and PIC is a viral mimic. Untreated cells from each fish served as parallel controls in the challenge study. The cells were further incubated for 24 h, harvested, treated with 600 μl RTL-Plus buffer (RNeasy®Plus kit Qiagen), centrifuged and frozen at −80 °C.
2.3. RNA extraction and qPCR
The impact of dietary Se levels and sources on the transcriptional changes in oxidative, immune and inflammatory response markersin vivo(liver and head kidney tissue) andin vitro(primary cultures of liver cells and head kidney leucocytes) were studied. The procedure for RNA extraction, reverse transcription and qPCR followed were as described in (Stenberg et al., 2019), except for the reference genes used to nor- malise the expression of target genes using the program geNorm version 3.5. The reference genes used for liver tissue were EF1a and β-actin;
while RPL and β-actin were used in liver cells. In the head kidney, it was β-actin and EF1a for both tissue and isolated cells. The details of the qPCR primers used for amplification of the reference and target genes are provided inTable 4.
2.4. Data analysis and statistics
The data from thein vivoandin vitrostudies were treated using tanks (n= 3) and individual fish (n= 3 to 5) as experimental units, respectively. Differences between groups were analysed using one-way or two-way ANOVA, following Tukey's multiple comparison test, for normally distributed datasets (Anderson-Darling normality test, p > .05). Non-normally distributed datasets (Anderson-Darling nor- mality test, p < .05) were analysed using Kruskal-Wallis non-para- metric test following Dunn's multiple comparison. Regression analysis was used to compare the efficacy of Se sources by slope-ratio method and dose response analysis to determine requirement estimates. The Table 4
Nucleotide Sequence of primers used in qPCR for mRNA expression analysis of target and house-keeping genes.
Gene Forward (5′- > 3′) Reverse (5′- > 3′) Ascension No. Reference
il1β GTATCCCATCACCCCATCAC GCAAGAAGTTGAGCAGGTCC NM_001123582 Stenberg et al. (2019)
p38mapk GGCACACAGACGATGAGATG ACAGCGTTCTGCCAGTGAG EF123661
5lox ACTAAGTTTGCTGCTTCGG CTGACTCCAGACCTCGTG CA387866
cox2 GGAGGCCTACTCCAACCTATT CGAACATGAGATTGGAACC AY848944
sod2 CCACGTCCATGCCTTTGG TCAGCTGCTGCAGTCACGTT DY718412
casp3 ACAGCAAAGAGCTAGAGGTCCAACAC AAAGCCAGGAGACTTTGACGCAG DQ008070 Martins et al. (2019)
cat CCAGATGTGGGCCGCTAACAA TCTGGCGCTCCTCCTCATTC Est04a09
viperin TCCTTGATGTTGGCGTGGAA GCATGTCAGCTTTGCTCCACA NM_001140939
bcl2 TGACAGATTTCATCTACGAGCGG GCCATCCAGCTCATCTCCAATC NM_001141086
tnfα GGCGAGCATACCACTCCTCT TCGGACTCAGCATCACCGTA AY848945
nfkβ CAGCGTCCTACCAGGCTAAAGAGAT GCTGTTCGATCCATCCGCACTAT CA341859 Holen et al. (2015)
pparα TCCTGGTGGCCTACGGATC CGTTGAATTTCATGGCGAACT DQ294237 Holen et al. (2014)
gpx2 TGTACCTCAAGGAGAAGCTGCCGT ATTAAGGCCATGGGATCGTCGC Est04e05 Todorčević et al. (2010)
gpx3 CCTTCCAGTACCTGGAGTTGAATGC CTCATGATTGTCTCCTGGCTCCTGT CA345853
aif AGGTGGAGTCCCAAGGAATCTGC CCCCAAGAAACCTCCTCCAATG TC37490 Kjær et al. (2016)
β-actin CCAAAGCCAACAGGGAGAA AGGGACAACACTGCCTGGAT BG933897 Stenberg et al. (2019)
ef1α TGCCCCTCCAGGATGTCTAC CAGCGTGATAGACTCGTTGC AF321836
rpl13 CCAATGTACAGCGCCTGAAA CGTGGCCATCTTGAGTTCCT NM_001141291 Hevrøy et al. (2015)
il-1b, interleukin 1 beta; p38mapk, p38 mitogen-activated protein kinases; 5lox, Arachidonate 5-lipoxygenase; cox2, cyclooxygenase2; sod2, superoxide dismutase2;
casp3, caspase 3; cat, catalase; viperin, virus inhibitory protein, endoplasmic reticulum [ER]-associated, interferon [IN]-inducible; bcl2, B-cell lymphoma 2; tnfα, tumor necrosis factor alpha; nfkβ, nuclear factor kappa-beta; pparα, peroxisome proliferator activated receptor alpha; gpx2, glutathione peroxidase 2; gpx3, glu- tathione peroxidase 3; aif, apoptosis-inducing factor; β-actin, beta actin; ef1α, elogation factor1 alpha; rpl13, ribosomal protein L13.
Table 5
Growth and zoo-technical indices of Atlantic salmon post-smolt fed different dietary levels and sources of selenium (Se) for 8 weeks.
Se source Se inclusion level
(mg kg−1diet) Analysed Se level in
diets (mg kg−1diet) Final body
weight (g) Weight
gain (g) Feed intake
(% day−1) Feed conversion
ratio (FCR) Specific growth rate (SGR)
Hepato- somatic index (HSI)
Hematocrit (Hct)
Basal diet 0 0.24 526.6 310.8 1.15 0.75 1.42 1.16 41.9ab
SS (Na2SeO3) 0.15 0.34 533.0 315.6 1.15 0.76 1.42 1.15 39.9a
0.4 0.65 530.2 313.8 1.16 0.75 1.43 1.16 44.6b
0.7 0.87 516.3 300.7 1.14 0.76 1.40 1.16 42.0ab
1.1 1.4 523.9 308.9 1.16 0.76 1.41 1.17 42.4ab
SM (Se-Met) 0.15 0.38 524.5 309.4 1.14 0.75 1.41 1.17 42.3ab
0.4 0.63 534.2 318.9 1.14 0.73 1.44 1.17 44.5b
pSD 14.9 14.3 0.03 0.01 0.04 0.03 1.1
p-value ns ns ns ns ns ns 0.02
Data presented as mean and pooled standard deviation (n = 3). Treatments effects were considered significant when p < .05 upon ANOVA followed by Tukey's post- hoc multiple comparison analysis. Different superscripts within a column indicate statistically significant difference between the groups; ns, treatment effects not significant,p > .05.
Fig. 1.Selenium (Se) concentration in Atlantic salmon post-smolt whole body (Fig.1A), liver (Fig. 1B), muscle (Fig. 1C), plasma (Fig. 1D), kidney (Fig. 1E) and liver/kidney ratio (Fig. 1F). Dark cir- cles and regression line represent sodium selenite (SS) fed groups, including the basal diet; open circles represent L-selenomethionine groups (SM). Each data point is presented as mean ± SD (n=3), the error bars are not visible for some data points. Test of significance between groups were analysed using one-way ANOVA following Tukey’s multiple com- parison test; groups with different letters within each graph differ significantly. The line represents the linear or non-linear regression of the five groups (basal + 4 SS groups). The regression equations were, whole body (Y = 0.14X + 0.09), plasma (Y = 0.05X2 + 0.18X – 0.05), muscle (Y = 0.06X + 0.08), liver (Y = 3.9X – 0.06), kidney (Y = 0.26X2 + 1.29X – 0.5) and liver/kidney ratio (Y = 1.9X2- 0.4X + 2.1). The response of Se concentration in kidney and liver/kidney ratio plateaued or not sig- nificantly different at or beyond 0.65 mg Se kg-1diet, respectively.
best fit of the data moving from simple to more complex relationships, was used. In figures with first order polynomials, p indicates the probability that the slope is equal to 0. In figures with second order fits, p gives the probability that the first order equation fits the data better than the second order equation. All the data analysis was performed in GraphPad Prism (ver. 8.05).
3. Results
3.1. Growth and zoo-technical indices
Atlantic salmon post-smolt fed the different experimental diets grew more than twice during the experimental period from a mean initial body weight of 216 ± 27 g to reach 527 ± 72 g. The weight gain between treatments was not significantly different, neither was there an impact of the experimental diets on feed intake, feed conversion ratio, specific growth rate or hepato-somatic index (Table 5,p > .05). He- matocrit percentage was significantly higher in fish fed the 0.4 mg Se supplemented feeds, irrespective of the source; the hematocrit of fish fed 0.15 SS diet was significantly lower (Table 5,p= .02).
3.2. Proximate and mineral composition of body
The chemical composition of the whole fish were not affected by the dietary treatments. The mean dry matter, protein, total fat, ash and gross energy content of the fish across all groups were 36.3 ± 0.4%,
17.5 ± 0.5%, 16.1 ± 0.5%, 1.9 ± 0.1% and 10.3 ± 0.2 kJ/g.
3.3. Whole body and tissue selenium concentrations
The Se concentrations in the whole body and tissues of Atlantic salmon post-smolt were significantly affected by the levels and sources of Se (Fig. 1). The Se concentration in whole body (Fig. 1A, p < .0001), liver (Fig. 1B, p < .0001) and muscle (Fig. 1C, p < .0001) increased linearly with increasing dietary Se concentra- tion. On the other hand, Se concentrations in plasma (Fig. 1D, p < .0001), kidney (Fig. 1E, p < .0001) and liver/kidney Se ratio (Fig. 1E, p < .0001) showed non-linear response, reaching saturation at or beyond dietary total Se (as SS) of 0.65 mg Se kg−1feed. Slope- ratio analysis was performed to assess the relative efficiency of the two Se sources in improving the Se status of the whole body and tissues (Table 6). The efficiency (measured as slope ratio) of SM was sig- nificantly higher than SS in the whole body (2.16, p < .0001), muscle (9.74, p < .0001) and plasma Se concentrations (2.49, p < .0001).
The contrary was significant (SM < SS,p < .01) in the liver (0.64, p < .01), liver/kidney ratio (0.18, p < .001) and faeces (0.27, p < .001); whereas, no significant difference was found in the slope ratio of hematocrit (0.85, p > .05) and kidney Se levels (0.95, p > .05). Dietary Se requirement of Atlantic salmon post smolt (on available basis) and the corresponding dietary Se levels (on total basis) estimated to meet the requireemnt as per these response parameters are provided inTable 7..
Table 6
Slope-ratio comparison on the efficacy of sodium selenite (SS) and selenomethionine (SM) based on different response criteria assessed.
Response Regression eq. (Y = bX + a)¥ Slope ratio (SM/SS) P-value (SMvsSS)
Selenomethionine (SM) Sodium selenite (SS)
Whole body Se y = 0.34× + 0.04 y = 0.16× + 0.08 2.16 < 0.0001
Plasma Se y = 0.38× – 0.004 y = 0.15× + 0.05 2.49 < 0.0001
Fillet Se y = 0.37× + 0.005 y = 0.04× + 0.09 9.74 < 0.0001
Hematocrit y = 7.0× + 40.0 y = 8.6× + 38.6 0.81 ns
Liver Se y = 1.9× + 0.6 y = 2.9× + 0.33 0.64 < 0.01
Kidney Se y = 0.83× + 0.33 y = 0.87× + 0.33 0.95 ns
Liver/kidney Se y = 0.23× + 1.92 y = 1.29× + 1.65 0.18 0.01
Faeces Se y = 0.06× + 0.02 y = 0.24× – 0.03 0.27 < 0.0001
¥ Linear regression (b, slope; a, intercept) of the response to three graded dietary Se levels from each Se source representing diet groups A, B and C for sodium selenite, and diets A, F and G for selenomethionine. The test of significance (at 0.05 level) to determine if the slope of the two regressions were significantly different was performed using GraphPad prism 8. ns, not statistically significant (p > .05).
Table 7
Dietary selenium (Se) to reach saturation in response criteria, Se requirement of Atlantic salmon post-smolt and the contribution of each diet tested in meeting the requirement.
Response
criteria X: Dietary total Se (as SS) to reach saturation or homeostasis(estimated), mg kg−1diet
Y: Respective Se requirement of Atlantic salmon(calculated), mg kg−1 diet
Z: Available Se supply (mg kg−1diet); and proportion of Se requirement met by each of the experimental diets(calculated), (%)
A B C D E F G
ANOVA1 Regression2 0.14 0.19 0.27 0.24 0.32 0.24 0.46
Body Se
balance – 0.66 0.27 51 68 99 88 115 89 166
Plasma Se 0.65 – 0.27 52 69 100 89 117 90 169
Hematocrit 0.65 0.65 0.27 54 71 103 92 120 93 174
Kidney Se 0.65 0.81 0.22–0.27 52–63 69–83 100–120 89–108 117–141 90–109 169–203
Liver/kidney
ratio 0.65 0.89 0.25–0.27 52–57 69–75 100–110 89–98 117–129 90–99 169–185
AAC 0.65 0.92 0.25–0.27 52–55 69–73 100–106 89–96 117–124 90–96 169–179
Faecal loss 0.65 0.93 0.26–0.27 52–55 69–72 100–105 89–94 117–123 90–95 169–177
Non-faecal loss 0.65 0.87 0.25–0.27 52–55 69–73 100–106 89–96 117–124 90–96 169–179
X, estimated values obtained using the five treatments (basal +4 SS supplemented groups);1based on post-hoc difference between treatments in respective response criteria;2parameter estimate from regression analysis. Y, calculated using the formula X*AAC%, the AAC values from diet respective to the Se level 42% for 0.65 or 27%, 0.8–0.9. Z, available Se level in diets A-G calculated as total Se concentration in diet*AAC% of the respective diets; proportion of Se requirement contributed by each experimental diet was calculated as follows, Z/Y*100.
3.4. Apparent availability, retention and loss of selenium
Data on apparent availability and mass balance of dietary Se is presented inFig. 2. The apparent availability of Se in the basal diet was 58.7%, which declined with SS supplementation and reached the lowest (26–27%) in the two highest SS supplemented groups (Fig. 2A;
p < .0001). On the contrary, the apparent availability of Se increased up to 68–72% in the two SM supplemented groups. Retention of dietary Se was superior and Se loss was less in SM fed fish (Fig. 2B,p< .0001).
The retention of dietary Se decreased linearly from 38% in the basal diet to 30% in the highest SS supplemented diet, while the retention increased up to 58–61% with Se supplementation as SM. Whole body Se balance as a function of dietary Se concentration indicated that fish fed SM were able to achieve whole body Se homeostasis with significantly lower Se supply than with SS (Fig. 2C, p < .0001). The dietary Se concentration to achieve body Se homeostasis with SS or SM were re- spectively 0.66 or 0.41 mg Se kg−1diet. Selenium loss, faecal (Fig. 2D, p < .0001) and non-faecal (Fig. 2E, p= .002) increased with Se supplementation as SS, but not with SM. The increase was non-linear in both cases and reached saturation at 0.65 mg Se kg−1diet; the broken- line regression estimates were 0.93 and 0.87 mg Se kg−1 diet,
respectively. Dietary Se requirement of Atlantic salmon post smolt (on available basis) and the corresponding dietary Se levels (on total basis) estimated to meet the requireemnt as per these response parameters are provided in Table 7. The present and projected Se load to the en- vironment is presented inFig. 2F. The Se load to the environment in the business-as-usual scenario of Norwegian aquaculture was estimated to be 0.89 t per year (in 2016), which can be potentially reduced to 0.65 or 0.26–0.34 t per year by managing Se supplementation as sodium selenite or selenomethionien, respectively, a reduction by 25 to 70%.
3.5. Glutathione redox homeostasis
The level of reduced glutathione (GSH), oxidised glutathione (GSSG), their ratio (GSH/GSSG) and the redox potential (Eh) were de- termined in the liver and head kidney (Fig. 3). Level of GSH in liver and head kidney were not affected by dietary Se (Fig. 3A). Liver GSSG data were fitted to a quadratic function (p= .04), indicating that the con- centrations were lower at intermediate SS supplementation of 0.4 and 0.65 mg Se kg−1(total dietary Se, 0.65 and 0.87 mg Se kg−1). A similar relationship between GSSG and SS supplementation was found in the head kidney (Fig. 3B). The liver GSH/GSSG ratio also followed a
0.0 0.5 1.0 1.5
0 20 40 60 80
AAC-Se(%)
SS SM
Diet Se (mg kg-1)
A
b b
ab
a a
bc c
0.0 0.5 1.0 1.5
0 20 40 60 80
Retention(%)
SS SM
Diet Se (mg kg-1)
B
a a
a a a
b b
0.5 1.0 1.5
-0.10 -0.05 0.00 0.05 0.10 0.15
BodySebalance
SS SM
Diet Se (mg kg-1)
C
0.0 0.5 1.0 1.5
0 20 40 60 80 100
FaecalSeloss(%)
SS SM
Diet Se (mg kg-1)
D
a a a a
b b b
0.5 1.0 1.5
-30 -20 -10 0 10 20
non-faecalSeloss(%)
SS SM Diet Se (mg kg-1)
E
a a
b
b b
ab b
Norway
(0.7 ppm) SS
(0.65 ppm) SM
(0.4 ppm) SM
(0.65 ppm)
-1.0 -0.5 0.0 0.5 1.0
Se(tons)
- 70% - 60%
- 25%
Retained in salmon Lost to the environment
F
0.25 ppm Se from ingredients 0.4 ppm Se from ingredients
Fig. 2.Apparent availability of selenium (Se) (Fig. 2A), retention (Fig. 2B), body Se balance (Fig. 2C), faecal (Fig. 2D) and non-faecal Se loss (Fig. 2E) in Atlantic salmon post-smolt fed different levels and sources of Se are presented here. Dark circles and regression line represent sodium selenite (SS) fed groups, including the basal diet; open circles and dashed line represent L-selenomethionine (SM) groups. Each data point is presented as mean ± SD (n=3), the error bars are not visible for some data points. Test of significance between groups were analysed using one-way ANOVA following Tukey’s multiple comparison test; groups with different let- ters within each graph differ significantly. The re- gression equations were, AAC of Se (Y = 77X2- 80X + 31), retention (Y = -7.2X + 40), faecal Se loss (Y
= 22X2+ 91X – 38) and non-faecal Se loss (Y = -37X2+ 78X – 35). In the body Se balance (SS, Y = 0.14X – 0.09; SM, Y = 0.34X – 0.14), X-intercept indicates the dietary Se concentration at which Se homeostasis was achieved with reference to the in- itial Se status of the fish (SS, 0.66; SM, 0.41 mg kg-1 diet). InFig. 2F, the retention (black bars) and loss (white bars) of dietary Se (in tons) estimated from salmon produced and feed used in Norwegian salmon farming (Aas et al., 2019) and Se retention obtained data in this study. The dietary Se levels for Norway was obtained from the Norwegian feed surveillance program, mean dietary Se content of 0.7 mg kg-1diet (Sanden et al., 2017). Compared to the estimated Se loss in Norwegian Atlantic salmon farming at present, the three proposed scenarios of dietary Se levels/source from this study had the potential to reduce Se loss to the environment by 25, 70 and 60%, respectively.
quadratic relationship with dietary Se, being maximum at intermediate dietary SS. The GSH/GSSG ratio in the head kidney was not affected by dietary Se (Fig. 3C). The redox potential (Eh) in the liver and kidney were not affected by the diets. The liver had higher levels of GSH and GSSG compared to head kidney (Fig. 3A, p < .0001; Fig. 3B, p < .0001). On the contrary, the GSH/GSSG ratio and redox potential were significantly higher in the head kidney compared to the liver (Fig. 3C,p< .02; 3D, p < .0001). The GSH, GSSG and redox potential of liver and kidney of fish fed the SM diets were similar to those fed the SS diets.
3.6. Dietary selenium induced apoptosis and inflammatory response in vivo The mRNA expression of genes involved in oxidative, apoptotic and inflammatory pathways in the liver and head kidney tissues were stu- died (Fig. 4). The gene expressions of the treatment groups were ana- lysed in comparison to the basal diet (control group). In the liver, Se supplementation at 0.4 ppm as SM significantly supressedaifexpression (4A,p= .002). The expression ofbcl2(Fig. 4B, p = .04) andcasp3 (Fig. 4C,p= .008) in 0.4 SS was significantly supressed in comparison to the basal diet group. Expression ofp38-MAPKwas not affected by dietary Se (Fig. 4D,p= .49). The expression ofgpx2was significantly lower in 0.4 SS and 0.15 SM group when compared to basal diet (Fig. 4E, p = .008). The expression ofsod2was unaffected by dietary Se (Fig. 4F,p = .09). In the head kidney, expression ofnfkβ (Fig. 4a, p = .002) and5lox(Fig. 4b, p = .04) were significantly higher, while tnfα(Fig. 4c,p= .06) showed a tendency to be higher in 0.15 SS fed fish compared to the basal diet group. Compared to the basal diet group, the expression oftlr7(Fig. 4d, p = .09) showed a tendency to be higher andcox2(Fig. 4e, p = .06) had the contrary in 0.15 SS group.
The expression ofgpx2was higher in 0.15 SS group compared to the basal diet group (Fig. 4f,p= .01).
3.7. Apoptosis, inflammatory and antioxidant response in vitro 3.7.1. Liver cells
The liver cells were affected by the impact of dietary Se, but not by the treatment of hydrogen peroxidein vitro(Fig. 5). Liver cells isolated from fish fed the 0.4 SM diet had significantly downregulated mRNA
expression ofaif(Fig. 5A,p= .001) andp38mapk(Fig. 5B,p= .0.001) compared to the basal diet group. On the other hand, the expression of cox2(Fig. 5C,p= .02, Kruskal-Wallis) andgpx3 (Fig. 5D, p = .04, Kruskal-Wallis) were significantly higher in 0.15 SS fed fish compared to the fish fed the basal diet. The expression ofbcl2, cat, sod2 and casp3 were not significantly affected by the treatments (Fig. 5E-H;p > .05).
3.7.2. Head kidney leukocytes
Gene expression of markers in inflammatory, apoptotic and oxida- tive response pathway were significantly altered through dietary Se and LPS or PIC exposure to the HKL (Fig. 6). Exposing the cells to LPS significantly induced the expression ofil-1βin all the groups except 0.4 SM (Fig. 6A). In cells exposed to LPS, Se supplied as SS tended to up- regulatecox2and5loxexpression, whereas Se supplied as SM tended to down regulate the expression of these genes. Consequently,cox2and 5lox expressions in 0.4 SS were significantly higher than in 0.4 SM (Fig. 6B and C). Exposing the cells to PIC significantly supressed the expression ofnfkβ in all the groups except 0.4 SM (Fig. 6D). The ex- pression of viperin was induced by PIC exposure but unaffected by dietary Se (Fig. 6E). Significant upregulation ofpparαexpression was observed in 0.4 SS and 0.15 SM group compared to the basal diet (Fig. 6F). The expression of gpx2 and cat were respectively down- regulated in 0.4 SM and 0.15 SS compared to the basal diet, no effect of LPS or PIC was observed (Fig. 6G and H). An interaction effect for PIC indiced cat upregulation was observed in 0.4 SS group. The expression ofaifwas downregulated in 0.15 SS and 0.4 SM compared to the basal diet; no effect of LPS or PIC observed (Fig. 6I).
4. Discussion
Atlantic salmon, despite being a major aquaculture species, the re- quirement for dietary Se is unknown (NRC, 2011). The optimal dietary Se level in low fish meal diets for Atlantic salmon post smolt estimated using a wide array of response criteria ranged between 0.65 and 0.93 mg kg−1diet, confirming our earlier estimate using a multi-nu- trient regression design (Antony Jesu Prabhu et al., 2019). Besides, the estimate obtained using SS as the Se source here compared well with most other sea water fish species like grouper, 0.7 and 0.9 mg kg−1diet (Lin, 2014;Lin and Shiau, 2005); cobia, 0.8 mg kg−1diet (Liu et al., Fig. 3.Circle and dark line, liver (L); square and dotted line, head kidney (HK). Filled symbols, sodium selenite (SS) in- cluding the basal diet and open symbols, selenomethionine (SM). Reduced glutathione (GSH, Fig. 3A), oxidised glu- tathione (GSSG, Fig. 3B), GSH/GSSG ratio (Fig. 3C) and glutathione redox potential (Eh,Fig. 3D). Each data point is presented as mean ± SD (n=3). Regression analyses was performed and none of the slopes of the linear models were different from zero. P-values indicate differences between organs and the probability that GSSG and GSH/GSSG models are linear instead of quadratic. Tissue comparison was per- formed using paired students’ t-test (n=21).
(caption on next page)
2010); black seabream, 0.86 mg kg−1 diet (Wang et al., 2019) and 0.94 mg kg−1diet for gilthead seabream (Domínguez et al., 2019),. The estimates discussed above are on total Se basis as very few Se re- quirement studies in fish provide data on Se availability. The Se re- quirement of Atlantic salmon on available basis (Table 7:ca.0.22 to 0.27 mg Se kg−1) is comparable with rainbow trout (Hilton et al., 1980), channel catfish (Gatlin and Wilson, 1984) and a predicted re- lationship between the selenocysteine content of SEPP1 and Se re- quirements in vertebrates (Penglase et al., 2015). Therefore, variations in optimal dietary Se levels for different fish species is largely driven by changes in bioavailability (Antony Jesu Prabhu et al., 2016).
Bioavailability of a nutrient includes absorption, retention, and utilization (Fairweather-Tait and Southon, 2003). The better absorption of SM relative to SS at the GIT of Atlantic salmon post-smolt was evi- dent through the AAC of Se, faeces Se concentration, plasma Se status, and the respective slope ratios. The concentration of Se excreted in the faeces was 4-fold higher in SS fed fish compared to those fed SM. So- dium selenite is absorbed in its selenite form (HSeO−) and the process is dependent on intracellular glutathione levels (Misra et al., 2012).
Anti-nutrient factors and other dietary additives can therefore interfere with selenite absorption at the GIT. Phytate-P and di‑calcium phosphate reduced apparent availability and absorption of Se when suppled as SS (Antony Jesu Prabhu et al., 2014;Silva et al., 2019). On the contrary, bothin vitrosolubility and apparent availabilityin vivowere higher with SM compared to SS and not affected by phytate-P in the diet (Silva et al., 2019;Silva et al., 2020). Once absorbed, Se is transported in the plasma (Hilton et al., 1982) and hence post-prandial plasma Se levels are used to assess Se absorption and/or status (Antony Jesu Prabhu et al., 2014; Bell and Cowey, 1989;Hambidge, 2003;Lin and Shiau, 2005). In the present study, SM improved the plasma Se status more effectively than SS. Fish fed SM supplemented diets required only half the total Se in the diets to achieve similar plasma Se than those fed SS diets. For instance, plasma Se status in fish fed 0.36 mg Se kg−1with SM was statistically similar with that of fish fed 0.65 mg Se kg−1with SS and the same was true for fish from 0.65 SMvs.1.4 SS. Moreover, the slope of plasma Se with SM supplementation was 2.6-fold higher compared to SS (SS, 0.15vs.SM, 0.38;p < .0001). This corresponds well with the overall assessment that Se requirement was satisfied at 0.65 and 0.4 mg kg−1of total dietary Se with Se supplementation as 0.4 ppm as SS and 0.17 ppm as SM, respectively.
Like mammals, liver and kidney are the major organs regulating Se metabolism in fish (Hilton et al., 1980;Hodson and Hilton, 1983). The selenium transported in the plasma is metabolised by liver, synthesised to selenoproteins, transported to extra-hepatic tissues and excreted by the kidney (Regina, 2015). The liver: kidney Se ratio (L:K) is thereby a useful measure to assess Se requirement and tolerance (Hilton et al., 1982). A steep increase in the L:K ratio signified excess dietary Se supply in rainbow trout (Hilton et al., 1982) and black seabream (Lee et al., 2008). Liver Se concentration increased linearly, while kindly Se levels plateaued at higher Se inclusion. Consequently, the L:K was significantly higher in fish fed 1.4 mg Se kg−1diet, which falls in the maximum tolerance range for selenite supplementation in plant based Atlantic salmon feeds (Berntssen et al., 2018). The metabolism and homeostasis of Se is not only differentially regulated by dietary Se le- vels, but also by Se sources (Janz, 2012). The regulated and unregulated Se pools bring about tissue specific differential absorption and retention of selenite and SM (Burk and Hill, 2015). Selenite and SM are the main
contributors to the regulated and unregulated Se pool, respectively. SM may also contribute to the regulated Se pool through transsulfuration pathway (Burk and Hill, 2015). The major non-specific Se pool is lo- calized in the muscle, which understandably exhibited significantly higher retention with SM in many fish species (Cotter et al., 2008;
Fontagne-Dicharry et al., 2015;Lin, 2014;Mechlaoui et al., 2019;Rider et al., 2010;Sele et al., 2018;Wischhusen et al., 2019). In contrast, liver Se concentration was higher with SS than SM; while kidney showed no source specific difference in Se deposition (Berntssen et al., 2018;Rider et al., 2010;Wischhusen et al., 2019). Atlantic salmon fed SM supple- mented feeds attained body Se homeostasis at a lower dietary Se level due to higher availability and tissue deposition of SM over SS, as in channel catfish (Wang and Lovell, 1997). The markers of specific Se pool (eg. SEPP1, glutathione peroxidase) are considered to be more robust indicators of Se status than those associated with the non-spe- cific Se pool (eg. muscle Se), both in mammals (Burk and Hill, 2015) and in fish (Antony Jesu Prabhu et al., 2016;Penglase et al., 2015).
Thereby, Se balance and slope ratio analysis in SM group might be biased and driven by increased Se retention in the non-specific Se pool, particularly the muscle. Nevertheless, this precludes the possible health benefits that can be obtained from the Se stored in the non-specific Se pool during challenging conditions.
Tissue glutathione concentrations and redox status are good in- dicators of Se requirement and toxicity in salmonids (Berntssen et al., 2018;Fontagne-Dicharry et al., 2015;Hamre et al., 2010). Glutathione (reduced, GSH) concentration in the liver of Atlantic salmon was 2-fold higher than in head kidney and was not affected by dietary Se, as ob- served with other dietary pro- or anti-oxidants (Hamre et al., 2010).
The concentration of GSH in a cell is high (i.e. 1–10 mM) and con- stitute > 98% of the total cellular glutathione pool. It is in steady state with the oxidised form, GSSG and their ratio expressed as the GSH/
GSSG ratio or the reduction potential indicates the redox status of the cell. GSSG had a quadratic relation with increasing SS supplementation in both liver and head kidney. The liver GSSG tended to be higher in low and high SS diets, indicating that the tissues were slightly more oxidised at these supplementation levels. The GSSG level in liver and head kidney and the GSH/GSSG ratio in the liver had min and max levels, respectively, in the 0.4 to 0.6 SS supplemented groups (total Se, 0.65 to 0.87 mg kg−1diet). This range corresponds well with the es- timated optimal total Se range to satisfy the Se requirement of Atlantic salmon post smolt, based on hematocrit, tissue Se status and body Se homeostasis(Table 7). With regard to the impact of Se sources, both SS and SM lowered GSH levels in Atlantic salmon at Se concentrations of 15 mg kg−1(Berntssen et al., 2018). In rainbow trout fry, organic Se improved GSH and GSH/GSSG ratio compared to SS (Fontagne- Dicharry et al., 2015), while the same effect was not reproduced in a subsequent study (Wischhusen et al., 2019). Source specific differences in glutathione metabolism arising from differential metabolic pathway of selenite or selenomethionine is suggested (Berntssen et al., 2017;
Burk and Hill, 2015).
Primary cultures of Atlantic salmon liver cells and HK leukocytes are well establishedin vitromodels to study tissue specific cellular re- sponses to nutrient depletion or repletionin vivo(Espe and Holen, 2013;
Espe et al., 2015;Holen et al., 2011;Martins et al., 2019). Grass carp fed Se-yeastin vivoexhibited anti-inflammatory effect of Se by mod- ulating nfkb, p38mapk and tnfα signalling pathways (Zheng et al., 2018a;Zheng et al., 2018b). Similar anti-inflammatory responses were Fig. 4.SS, sodium selenite; SM, selenomethionine. Upper case alphabets represent liver tissue (Fig. 4A-F); lower case alphabets represent head kidney tissue (Fig. 4a- f). Liver:aif, apoptosis-inducing factor (Fig. 4A);bcl2, B-cell lymphoma 2 (Fig. 4B);casp3, caspase 3 (Fig. 4C);p38mapk, p38 mitogen-activated protein kinases (Fig. 4D);gpx2, glutathione peroxidase 2 (Fig. 4E);sod2, superoxide dismutase2 (Fig. 4F). Head kidney:nfkβ,Nuclear Factor kappa-beta (Fig. 4a);5lox, Arachidonate 5-lipoxygenase (Fig. 4b); tnfα, tumor necrosis factor alpha (Fig. 4c); trl7, toll-like receptor7 (Fig. 4d);cat, catalase;cox2, cyclooxygenase 2 (Fig. 4e) andgpx2, glutathione peroxidase 2 (Fig. 4f). Data presented as mean and pooled standard deviation (n=4). Treatments effects were considered significant when p < 0.05 upon ANOVA followed by Tukey’s post-hoc multiple comparison analysis of each treatment with that of the control (basal diet). Asterix symbol above a treatment group within a graph indicate statistically significant difference of the respective treatment group compared to the basal diet group (*, p < 0.05; **, p < 0.01).
(caption on next page)
reported in rainbow trout fed Se-yeast when subjected to intra-perito- nial injection of PICin vivo(Pacitti et al., 2016). In Atlantic salmon liver cells, suppression ofp38mapkexpression indicated protection against apoptosis (Espe and Holen, 2013). Se-deficiency and excess induced upregulation ofp38mapkin spleen, kidney and skin of grass carp was supressed by optimal Se-yeast supplementation at 0.6 ppm (Zheng et al., 2018a;Zheng et al., 2018b). In the present study, the in vitro expression ofp38mapkandaifin the liver cells were significantly re- duced in 0.4 SM group compared to the basal and 0.4 SS group. Excess selenium induced apoptotic and oxidative stress in liver cells is asso- ciated with hydropic degeneration and other histological changes (He et al., 2014). In gilthead seabream, SM supplementation above re- quirement protected liver and muscle tissue against oxidative damage (Mechlaoui et al., 2019), while excess SS inclusion resulted in hydropic degeneration of hepatocytes (Domínguez et al., 2019). Glutathione peroxidase expressionin vivois a potent anti-oxidant marker responsive
to dietary Se in rainbow trout (Fontagne-Dicharry et al., 2015). On the contrary, the reducedin vivoexpression ofgpx2by Se supplementation, irrespective of the source could be related to the decreased oxidation of glutathione in these groups. In the head kidney,in vivoexpression of nfkβ,5lox,tnfαandgpx2were all higher in 0.15 SS fed fish compared to the basal diet group. Whereas in HKL cellsin vitro, LPS or PIC activated induction ofil-1b,cox2,nfkb and gpx2were attenuated in fish fed 0.4 SM diets. Similar effects of Se deficiency or repletion were reported in RAW 264.7 macrophages (Zamamiri-Davis et al., 2002), mammary epithelial cells in primary culture (Zhang et al., 2014) and murine macrophage cultures (Kim et al., 2004). The attenuation effect of Se was observed only in 0.4 SM group but not in fish from 0.15 SM and 0.4 SS groups that satisfied the requirement for body Se homeostasis. This is consistent within vivo observations in rainbow trout (Pacitti et al., 2016), grass carp, (Zheng et al., 2018a,Zheng et al., 2018b), channel catfish (Wang et al., 1997) and hybrid striped bass (Jaramillo and Fig. 5.The black and white bars represent control and hydrogen peroxide (H2O2) treated cells.aif, apoptosis-inducing factor (Fig. 5A);p38mapk, p38 mitogen- activated protein kinases (Fig. 5B);cox2, cyclooxygenase 2 (Fig. 5C);gpx3, glutathione peroxidase 3 (Fig. 5D);bcl2, B-cell lymphoma 2 (Fig. 5E);cat, catalase (Fig. 5F);sod2, superoxide dismutase 2 (Fig. 5G);casp3, caspase 3 (Fig. 5H). Data presented as mean and standard deviation (n=5). The test of significance performed by two-way ANOVA with diet and stimulus as main factors, followed by Tukey’s multiple comparison of each treatment with that of the control (basal diet) for diet effect. Asterix symbol above a treatment group within a graph indicate statistically significant difference of the respective treatment group compared to the basal diet group (*, p < 0.05; **, p < 0.01).
Fig. 6.The black, white and grey bars represent control, lipopolysaccharide (LPS) and Polyinosinic:polycytidylic acid (PIC) treatments, respectively.il-1b, interleukin 1 beta (Fig. 6A);cox2, cyclooxygenase2 (Fig. 6B);5lox, Arachidonate 5-lipoxygenase (Fig. 6C);nfkβ, nuclear factor kappa-beta (Fig. 6D);viperin, virus inhibitory protein, endoplasmic reticulum [ER]-associated, interferon [IN]-inducible (Fig. 6E);pparα, peroxisome proliferator activated receptor alpha (Fig. 6F);gpx2, glu- tathione peroxidase 2 (Fig. 6G);cat, catalase (Fig. 6H);aif, apoptosis-inducing factor (Fig. 6I). Data presented as mean and standard deviation (n=3). The test of significance was performed by two-way ANOVA with diet and stimulus (LPS or PIC) as main factors, followed by Tukey’s multiple comparison for diet or stimulus effect based on the ANOVA results. The asterix superscripts (*, p < 0.05; **, p < 0.01) indicate statistically significant differences between respective dietary group and the basal diet. The dollar superscripts ($, p < 0.05; $$, p < 0.01) indicates the effect of the stimulus (LPS or PIC) compared to the untreated control within each diet group.
